The present invention relates to an inverse discrete-cosine transform apparatus for transforming input discrete cosine coefficients to inverse discrete-cosine coefficients.
An inverse discrete-cosine transform apparatus is incorporated into an image-decoding apparatus that is designed to decode compressed image data. In the image-decoding apparatus, the inverse discrete-cosine transform apparatus transforms image data provided in the form of discrete-cosine coefficients, into inverse discrete-cosine coefficients.
More precisely, the inverse discrete-cosine transform apparatus transforms input coefficients, in units of discrete-cosine blocks, thereby to generate image data. Each discrete-cosine block is, for example, an 8xc3x978 matrix that is composed of discrete-cosine coefficients arranged in rows and columns.
Discrete-cosine coefficients can be transformed to inverse discrete-cosine coefficients by applying the following equation (1) of inverse transform:                                           S            xy                    =                                    ∑                              u                =                D                            7                        ⁢                                          ∑                                  v                  =                  D                                7                            ⁢                                                C                  u                                ⁢                                  C                  v                                ⁢                                  D                  uv                                ⁢                cos                ⁢                                                                            (                                                                        2                          ⁢                          x                                                +                        1                                            )                                        ⁢                    u                    ⁢                                          xe2x80x83                                        ⁢                    π                                    16                                ⁢                cos                ⁢                                  xe2x80x83                                ⁢                                                                            (                                                                        2                          ⁢                          y                                                +                        1                                            )                                        ⁢                    v                    ⁢                                          xe2x80x83                                        ⁢                    π                                    16                                                                    ⁢                  
                ⁢                  {                                                                      Cu                  =                                      Cv                    =                                                                                            1                                                      2                                                                          ⁢                        u                                            =                                              v                        =                        0                                                                                                                                                                                      Cu                    =                                          Cv                      =                                              1                        ⁢                        u                                                                              ,                  v1                  ,                                      2                    ⁢                                          xe2x80x83                                        ⁢                    …                                    ⁢                                      xe2x80x83                                    ,                  7                                                                                        (        1        )            
where Duv is the discrete-cosine coefficients, i.e., the elements of a discrete-cosine block, Sxy is pixel data. In the symbol Duv, and v indicate the horizontal component and vertical component of the discrete-cosine block, respectively. Similarly, in the symbol, x and y indicate the horizontal component and vertical component of the pixel data, respectively.
As seen from the equation (1), the inverse discrete-cosine transform can be accomplished by performing matrix calculus on discrete-cosine coefficients and inverse discrete-cosine coefficients. Hence, the inverse discrete-cosine transform apparatus may have a matrix algebraic circuit that comprises multipliers and adders. In this case, the apparatus can effect inverse discrete-cosine transform on an input image of standard resolution or high resolution, which has been subjected to discrete-cosine transform, thereby to generate image data that has the same resolution as the input image.
To provide such a matrix algebraic circuit, various methods have been devised. Each method is designed to reduce the number of operations that the matrix algebraic circuit needs to perform. In November 1984 Mr. Beyong Gi Lee published a fast cosine transform (FCT) algorithm in IEEE Transaction on Acoustics, Speech and Signal Processing, Vol. 32, No. 6, pp. 1243. This algorithm describes a method of reducing the number of necessary operations. A circuit, designed totally on the basis of the algorithm, has been developed.
Thus, a fast algorithm optimal for an inverse discrete-cosine transform of discrete-cosine blocks of a specific size, for example 8xc3x978 inverse discrete-cosine blocks, may be formulated and applied. Then, it is possible to provide a small, high-speed matrix algebraic circuit.
An inverse discrete-cosine transform apparatus is known which converts a high-resolution image subjected to discrete-cosine transform, to an image having standard resolution. That is, the apparatus accomplishes compression inverse discrete-cosine transform. Japanese Patent Application Publication No. 2000-041261 discloses an inverse discrete-cosine transform apparatus of this type.
Compression inverse discrete-cosine transform may be performed on a discrete-cosine block subjected to discrete-cosine transform in field discrete-cosine mode, thereby providing first pixel data. Further, compression inverse discrete-cosine transform may be carried out on a discrete-cosine block subjected to discrete-cosine transform in frame discrete-cosine mode, thereby providing second pixel data. The first pixel data and the second pixel data, thus provided, inevitably have a phase difference in the vertical direction. If an image-decoding apparatus incorporates an inverse discrete-cosine transform apparatus that effects the same compression inverse discrete-cosine transform on these two discrete-cosine blocks of different types, the quality of the image the apparatus outputs will deteriorated.
In order to eliminate the phase difference in the vertical direction, two types of compression inverse discrete-cosine transform apparatuses have been invented. The first type is a field-mode, compression, inverse discrete-cosine transform apparatus that performs compression inverse discrete-cosine transform on a discrete-cosine block subjected to discrete-cosine transform in field discrete-cosine mode. The second type is a frame-mode, compression, inverse discrete-cosine transform apparatus that divides a discrete-cosine block subjected to discrete-cosine transform in frame discrete-cosine mode, into fields, thereby to accomplish the compression inverse discrete-cosine transform on the discrete-cosine block.
The field-mode, compression, inverse discrete-cosine transform apparatus will be described first, which performs compression inverse discrete-cosine transform on a discrete-cosine block subjected to discrete-cosine transform in field discrete-cosine mode.
The field-mode, compression, inverse discrete-cosine transform apparatus receives an 8xc3x978 discrete-cosine block input in the form of a bit stream. The apparatus then performs inverse discrete-cosine transform on only the lower 4xc3x974 coefficients of the 8xc3x978 discrete-cosine block. In other words, the apparatus performs compression inverse discrete-cosine transform on the basis of four lower points existing in a lower region with respect to both the horizontal and the vertical direction. The field-mode, compression, inverse discrete-cosine transform apparatus can convert one discrete-cosine block to 4xc3x974 pixel data as it carries out the compression inverse discrete-cosine transform.
It will be described how the frame-mode, compression, inverse discrete-cosine transform apparatus divides a discrete-cosine block subjected to discrete-cosine transform in frame discrete-cosine mode, into fields, thereby to accomplish compression inverse discrete-cosine transform on the discrete-cosine block.
As shown in FIG. 1, the frame-mode, compression, inverse discrete-cosine transform apparatus receives a bit stream that has been generated by compressing and encoding a high-resolution image. The bit stream is input to the apparatus, in the form of a discrete-cosine block.
First, in Step S1, the apparatus performs 8xc3x978 inverse discrete-cosine transform on the discrete-cosine coefficients y of the discrete-cosine block. (Of all discrete-cosine coefficients of the block, only those in the vertical direction are shown as y1 to y8 in FIG. 1.) As a result, 8xc3x978 pixel data x is decoded. (Of all pixel data items of the block, only those in the vertical direction are shown as items x1 to x8 in FIG. 1.)
In Step S2, the pixel data items are alternately extracted in the vertical direction, thus dividing the 8xc3x978 pixel data into a 4xc3x974 top-field pixel block and a 4xc3x974 bottom-field pixel block, which correspond to pixel blocks obtained by interlaced scanning. More specifically, pixel data items x1, x3, x5 and X7 for the first, third, fifth and seventh horizontal lines, respectively, are extracted and combined, thus forming a pixel block that corresponds to a top field. Pixel data items X2, X4, X6 and x8 for the second, fourth, sixth and eighth horizontal lines, respectively, are extracted and combined, forming a pixel block that corresponds to a bottom field. This process of dividing the pixels of a discrete-cosine block into two pixel blocks that correspond to interlaced-scan pixel blocks is called xe2x80x9cfield divisionxe2x80x9d (also known as xe2x80x9cfield separationxe2x80x9d).
In Step S3, the apparatus carries out 4xc3x974 discrete-cosine transform (DCT4xc3x974) on the two pixel blocks that have been generated by means of field division.
In Step S4, the apparatus extracts the higher ones of the discrete-cosine coefficients z for the top-field pixel block generated by effecting the 4xc3x974 discrete-cosine transform. (Of all coefficients of the top-field pixel block, only the discrete-cosine coefficients in the vertical direction are shown as z1, Z3, Z5 and Z7 in FIG. 1.) The higher discrete-cosine coefficients extracted are combined, forming a pixel block composed of 2xc3x972 discrete-cosine coefficients. Also, the apparatus extracts the higher ones of the discrete-cosine coefficients z for the bottom-field pixel block generated by effecting the 4xc3x974 discrete-cosine transform. (Of all coefficients of the bottom-field pixel block, only the discrete-cosine coefficients in the vertical direction are shown as Z2, Z4, Z6 and Z8 in FIG. 1.) The higher discrete-cosine coefficients extracted are combined, forming a pixel block composed of 2xc3x972 discrete-cosine coefficients.
In Step S5, the apparatus effectuates 2xc3x972 inverse discrete-cosine transform (IDCT 2xc3x972) on the pixel block composed of the higher discrete-cosine coefficients that have been extracted from the top-field pixel block. As a result, 2xc3x972 pixel data xxe2x80x2 is decoded. (Of all pixel data items of the top-field pixel block, only those in the vertical direction are shown as items xxe2x80x21 and Xxe2x80x23 in FIG. 1. Also, of all pixel data of the bottom-field pixel block, only the pixel data in the vertical direction are shown as items xxe2x80x22 and xxe2x80x24 in FIG. 1.)
In Step S6, the pixel data items of the top-field pixel block and the pixel data items of the bottom-field pixel block are alternately selected for each line and synthesized in the vertical direction, thus performing compression inverse discrete-cosine transform. A discrete-cosine block composed of 4xc3x974 pixel data items is thereby generated. This process of selecting the pixel data items of the top-field and bottom-field pixel blocks and combining them in the vertical direction shall be called xe2x80x9cframe synthesis.xe2x80x9d
Performing Steps S1 to S6, the frame-mode, compression, inverse discrete-cosine transform apparatus can generate pixel data that is in the same phase as the pixel data generated in the field-mode, compression, inverse discrete-cosine transform apparatus.
The frame-mode, compression, inverse discrete-cosine transform apparatus effects Steps S1 to S6 by using a single matrix. To state it more specifically, the apparatus executes matrix calculus on the matrix [FS] of the following equation (2) and the discrete-cosine coefficients y (Y1 to Y8) of a discrete-cosine block, generating the pixel data xxe2x80x2 (items xxe2x80x21 to xxe2x80x24) of a discrete-cosine block obtained by compression inverse discrete-cosine transform. Note that matrix [FS] has been formed as the above-mentioned process is carried out by applying addition theorem.                               [          FS          ]                =                              1                          2                                ⁡                      [                                                            A                                                  B                                                  D                                                                      -                    E                                                                    F                                                  G                                                  H                                                  I                                                                              A                                                                      -                    C                                                                                        -                    D                                                                    E                                                                      -                    F                                                                                        -                    G                                                                                        -                    H                                                                                        -                    J                                                                                                A                                                  C                                                                      -                    D                                                                                        -                    E                                                                                        -                    F                                                                    G                                                                      -                    H                                                                    J                                                                              A                                                                      -                    B                                                                    D                                                  E                                                  F                                                                      -                    G                                                                    H                                                                      -                    I                                                                        ]                                              (        2        )            
A to J in the equation (2) are as follows:       A    =                            1                      2                          ⁢                  xe2x80x83                ⁢        D            =                                    1            4                    ⁢                      xe2x80x83                    ⁢          H                =                              1            4                    +                      1                          2              ⁢                              2                                                              B    =                            COS          ⁢                      xe2x80x83                    ⁢                      π            16                          +                  COS          ⁢                      xe2x80x83                    ⁢                                    3              ⁢              π                        16                          +                  3          ⁢          COS          ⁢                      xe2x80x83                    ⁢                                    5              ⁢                              xe2x80x83                            ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    7              ⁢                              xe2x80x83                            ⁢              π                        16                              4            E    =                            COS          ⁢                      xe2x80x83                    ⁢                      π            16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    3              ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    5              ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    7              ⁢              π                        16                              4            I    =                            COS          ⁢                      xe2x80x83                    ⁢                      π            16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    3              ⁢              π                        16                          +                  3          ⁢          COS          ⁢                      xe2x80x83                    ⁢                                    5              ⁢              π                        16                          +                  COS          ⁢                      xe2x80x83                    ⁢                                    7              ⁢              π                        16                              4            F    =                            COS          ⁢                      xe2x80x83                    ⁢                      π            8                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    3              ⁢              π                        8                              4            C    =                            COS          ⁢                      xe2x80x83                    ⁢                      π            16                          -                  3          ⁢          COS          ⁢                      xe2x80x83                    ⁢                                    3              ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    5              ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    7              ⁢              π                        16                              4                                    G          =                                                    COS                ⁢                                  xe2x80x83                                ⁢                                  π                  16                                            -                              COS                ⁢                                  xe2x80x83                                ⁢                                                      3                    ⁢                    π                                    16                                            +                              COS                ⁢                                  xe2x80x83                                ⁢                                                      5                    ⁢                    π                                    16                                            +                              COS                ⁢                                  xe2x80x83                                ⁢                                                      7                    ⁢                    π                                    16                                                      4                                                        J          =                                                    COS                ⁢                                  xe2x80x83                                ⁢                                  π                  16                                            +                              3                ⁢                COS                ⁢                                  xe2x80x83                                ⁢                                                      3                    ⁢                    π                                    16                                            -                              COS                ⁢                                  xe2x80x83                                ⁢                                                      5                    ⁢                    π                                    16                                            +                              COS                ⁢                                  xe2x80x83                                ⁢                                                      7                    ⁢                    π                                    16                                                      4                              
Fast algorithm may be used to effectuate the 4xc3x974 compression inverse discrete-cosine transform in the field-mode, compression, inverse discrete-cosine transform apparatus and to perform Steps S1 to S6, i.e., compression inverse discrete-cosine transform, in the frame-mode, compression, inverse discrete-cosine transform apparatus.
In both apparatuses, applying a fast algorithm can carry out the compression inverse discrete-cosine transform. An example of a fast algorithm is the Wang algorithm (see Zhong DE Wang., xe2x80x9cFast Algorithms for the Discrete W Transform and for the Discrete Fourier Transformxe2x80x9d, IEEE Tr. ASSP-32, No. 4, pp. 803-816, Aug. 1984).
The matrix representing the compression discrete-cosine transform that the field-mode, compression, inverse discrete-cosine transform apparatus executes can be decomposed as shown in the following equation (3), by applying the Wang algorithm:                                                                         [                                  C                  d                  II                                ]                                            -                1                                      ⁡                          [                              C                d                III                            ]                                =                                                                                                                1                                              2                                                              ⁡                                          [                                                                                                    1                                                                                0                                                                                0                                                                                1                                                                                                                                0                                                                                1                                                                                1                                                                                0                                                                                                                                0                                                                                1                                                                                                              -                              1                                                                                                            0                                                                                                                                1                                                                                0                                                                                0                                                                                                              -                              1                                                                                                                          ]                                                        ⁡                                      [                                                                                                                        [                                                          C                              2                              III                                                        ]                                                                                                                                xe2x80x83                                                                                                                                                                            xe2x80x83                                                                                                                                [                                                                                          C                                _                                                            2                              IV                                                        ]                                                                                                                ]                                                  ⁡                                  [                                                                                    1                                                                    0                                                                    0                                                                    1                                                                                                            0                                                                    0                                                                    1                                                                    0                                                                                                            0                                                                    0                                                                    0                                                                    1                                                                                                            0                                                                    1                                                                    0                                                                    0                                                                              ]                                            ⁢                              
                            [                              C                2                III                            ]                        =                                                            [                                      C                    d                    II                                    ]                                T                            =                                                [                                                                                                              1                                                      2                                                                                                                                                1                                                      2                                                                                                                                                                                                                    1                                                          2                                                                                -                                                                                                                      1                                                      2                                                                                                                                ]                                =                                                                                                    1                                                  2                                                                    ⁡                                              [                                                                                                            1                                                                                      1                                                                                                                                          1                                                                                                                      -                                1                                                                                                                                    ]                                                              ⁢                                          C                      r                                                        =                                      COS                    ⁡                                          (                                              r                        ⁢                                                  xe2x80x83                                                ⁢                        π                                            )                                                                                                          ⁢                  
                ⁢                                                                              [                                                            C                      _                                        2                    IV                                    ]                                =                                  [                                                                                                              -                                                      C                                                          1                              8                                                                                                                                                                            C                                                      9                            8                                                                                                                                                                                        C                                                      9                            8                                                                                                                                                C                                                      1                            8                                                                                                                                ]                                                                                                        =                                                                            [                                                                                                    1                                                                                0                                                                                                              -                              1                                                                                                                                                            0                                                                                1                                                                                1                                                                                              ]                                        ⁡                                          [                                                                                                                                                                  -                                                                  C                                                                      1                                    8                                                                                                                              +                                                              C                                                                  9                                  8                                                                                                                                                                          0                                                                                0                                                                                                                                0                                                                                                                                              C                                                                  1                                  8                                                                                            +                                                              C                                                                  9                                  8                                                                                                                                                                          0                                                                                                                                0                                                                                0                                                                                                              C                                                              9                                8                                                                                                                                                        ]                                                        ⁡                                      [                                                                                            1                                                                          0                                                                                                                      0                                                                          1                                                                                                                      1                                                                                                      -                            1                                                                                                                ]                                                                                                          (        3        )            
FIG. 2 is a flowchart explaining how the Wang algorithm is applied in the field-mode, compression, inverse discrete-cosine transform apparatus. As can be understood from the flowchart, five multipliers 14a to 14e and nine adders 14f to 14n are used to achieve a compression inverse discrete-cosine transform at high speed.
The Wang algorithm is applied, decomposing the matrix [FS] into one expressed by the following equation (4). Note that the matrix [FS] is processed by the frame-mode, compression, inverse discrete-cosine transform apparatus.                               [          FS          ]                =                                                                                                                        1                                              2                                                              ⁡                                          [                                                                                                    1                                                                                0                                                                                0                                                                                0                                                                                                                                0                                                                                0                                                                                0                                                                                1                                                                                                                                0                                                                                1                                                                                0                                                                                0                                                                                                                                0                                                                                0                                                                                1                                                                                0                                                                                              ]                                                        ⁡                                      [                                                                                            1                                                                          0                                                                          1                                                                          0                                                                                                                      0                                                                          1                                                                          0                                                                          1                                                                                                                      1                                                                          0                                                                                                      -                            1                                                                                                    0                                                                                                                      0                                                                          1                                                                          0                                                                                                      -                            1                                                                                                                ]                                                  ⁡                                  [                                                                                                              [                                                      M                            1                                                    ]                                                                                                                      xe2x80x83                                                                                                                                                              xe2x80x83                                                                                                                      [                                                      M                            2                                                    ]                                                                                                      ]                                            ⁡                              [                                                                            10000000                                                                                                  00100000                                                                                                  00001000                                                                                                  00000010                                                                                                  00010000                                                                                                  00000100                                                                                                  01000000                                                                                                  00000001                                                                      ]                                      ⁢                          
                        [                          M              1                        ]                    =                                                                                          [                                                                                            1                                                                          1                                                                                                                      1                                                                                                      -                            1                                                                                                                ]                                    ⁡                                      [                                                                                            1                                                                          0                                                                          0                                                                          0                                                                                                                      0                                                                          1                                                                          1                                                                          1                                                                                      ]                                                  ⁡                                  [                                                                                    A                                                                    0                                                                    0                                                                    0                                                                                                            0                                                                    D                                                                    0                                                                    0                                                                                                            0                                                                    0                                                                    F                                                                    0                                                                                                            0                                                                    0                                                                    0                                                                    H                                                                              ]                                            ⁢                              
                            [                              M                2                            ]                        =                                                            [                                                                                    1                                                                    1                                                                    0                                                                                                            1                                                                    0                                                                    1                                                                              ]                                ⁡                                  [                                                                                                              -                          1                                                                                            1                                                                    0                                                                    0                                                                    0                                                                    0                                                                                                            0                                                                    0                                                                    1                                                                    0                                                                    1                                                                    0                                                                                                            0                                                                    0                                                                    0                                                                    1                                                                    0                                                                    1                                                                              ]                                            ⁡                              [                                                                            E                                                              0                                                              0                                                              0                                                                                                  0                                                              G                                                              0                                                              0                                                                                                  0                                                              0                                                              B                                                              0                                                                                                  0                                                              0                                                              C                                                              0                                                                                                  0                                                              0                                                              0                                                              I                                                                                                  0                                                              0                                                              0                                                              J                                                                      ]                                                                        (        4        )            
A to J in the equation (4) are as follows:                                           A            =                          1                              2                                              ⁢                      xe2x80x83                                                F          =                                                    COS                ⁢                                  π                  8                                            -                              COS                ⁢                                  xe2x80x83                                ⁢                                                      3                    ⁢                                          xe2x80x83                                        ⁢                    π                                    8                                                      4                                                        D          =                      1            4                                                H          =                                    1              4                        +                          1                              2                ⁢                                  2                                                                              B    =                            COS          ⁢                      xe2x80x83                    ⁢                      π            16                          +                  COS          ⁢                      xe2x80x83                    ⁢                                    3              ⁢              π                        16                          +                  3          ⁢          COS          ⁢                      xe2x80x83                    ⁢                                    5              ⁢                              xe2x80x83                            ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    7              ⁢                              xe2x80x83                            ⁢              π                        16                              4            C    =                            COS          ⁢                      xe2x80x83                    ⁢                      π            16                          -                  3          ⁢          COS          ⁢                      xe2x80x83                    ⁢                                    3              ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    5              ⁢              π                        16                          -                  COS          ⁢                      xe2x80x83                    ⁢                                    7              ⁢              π                        16                              4                                                E            =                                                            COS                  ⁢                                      xe2x80x83                                    ⁢                                      π                    16                                                  -                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            3                      ⁢                      π                                        16                                                  -                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            5                      ⁢                      π                                        16                                                  -                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            7                      ⁢                      π                                        16                                                              4                                ⁢                      
                    ⁢                      G            =                                                            COS                  ⁢                                      xe2x80x83                                    ⁢                                      π                    16                                                  -                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            3                      ⁢                      π                                        16                                                  +                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            5                      ⁢                      π                                        16                                                  +                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            7                      ⁢                      π                                        16                                                              4                                                                                I            =                                                            COS                  ⁢                                      xe2x80x83                                    ⁢                                      π                    16                                                  -                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            3                      ⁢                      π                                        16                                                  +                                  3                  ⁢                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            5                      ⁢                      π                                        16                                                  +                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            7                      ⁢                      π                                        16                                                              4                                ⁢                      
                    ⁢                      J            =                                                            COS                  ⁢                                      xe2x80x83                                    ⁢                                      π                    16                                                  +                                  3                  ⁢                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            3                      ⁢                      π                                        16                                                  -                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            5                      ⁢                      π                                        16                                                  +                                  COS                  ⁢                                      xe2x80x83                                    ⁢                                                            7                      ⁢                      π                                        16                                                              4                                          
FIG. 3 is a flowchart explaining how the Wang algorithm is applied in the frame-mode, compression, inverse discrete-cosine transform apparatus. As seen from this flowchart, ten multipliers 15a to 15j and thirteen adders 15k to 15w are used to accomplish a compression inverse discrete-cosine transform at high speed.
Hitherto, the inverse discrete-cosine transform has been effected by three different methods. The first method performs inverse discrete-cosine transform on a high-resolution image or a standard-resolution image, either subjected to discrete-cosine transform, while maintaining the resolution of the image. (Hereinafter, the first method will be referred to as xe2x80x9cstandard inverse discrete-cosine transform.xe2x80x9d) The second method carries out inverse discrete-cosine transform on a high-resolution image subjected, converting the image to one having a reduced resolution. (Hereinafter, the second method will be called xe2x80x9ccompression, inverse discrete-cosine transform.xe2x80x9d) The third method effects field discrete-cosine transform on a discrete-cosine block subjected, thus dividing the block into fields. (Hereinafter, the third method will be referred to as xe2x80x9cfield-division, inverse discrete-cosine transform.xe2x80x9d) The inverse discrete-cosine transform apparatuses that perform these three methods, respectively, are dedicated hardware units.
Recently, image data is digitized. More and more apparatuses complying with the MPEG (Moving Picture Experts Group) system are used in broadcast stations and data-receiving sites such as households, for two reasons. First, the apparatuses perform orthogonal transformation and motion compensation on digital image data that has redundancy, thereby compressing the image data. Second, the image data can be transmitted and stored with higher efficiency than in the case it is not so compressed at all.
The image data that will be transmitted in digital broadcasting in increasing amounts contains both standard-resolution data and high-resolution data. The data-receiving apparatus that receives the image data needs to have an inverse discrete-cosine transform apparatus that can decode both the standard-resolution data and the high-resolution data.
To perform the above-mentioned different methods, however, a inverse discrete-cosine transform apparatus needs to have many multipliers and adders and will become complex, large and expensive. This is inevitably because the discrete-cosine blocks processed in the methods differ in size.
The present invention has been made in consideration of the foregoing. An object of the invention is to provide an inverse discrete-cosine transform apparatus that has a simple structure and can, nonetheless, perform both standard inverse discrete-cosine transform and compression, inverse discrete-cosine transform and field-division, and/or inverse discrete-cosine transform.
To achieve the object, an inverse discrete-cosine transform apparatus according to the invention is designed to perform inverse discrete-cosine transform on a discrete-cosine block that is a matrix composed of at most 8xc3x978 discrete-cosine coefficients. The apparatus comprises: eight discrete-cosine transform multipliers for multiplying the discrete-cosine coefficients input in the form of a bit stream, by coefficients; ten field, compression discrete-cosine transform multipliers for multiplying the discrete-cosine coefficients input in the form of a bit stream, by coefficients; eight selecting means for receiving the discrete-cosine coefficients multiplied by the coefficients in the eight discrete-cosine transform multipliers and the discrete-cosine coefficients multiplied by the coefficients in the ten field, compression discrete-cosine transform multipliers; control means for controlling the eight selecting means so that, when the discrete-cosine block is not subjected to field division, one of the values input from the eight discrete-cosine transform multipliers to the eight selecting means may be selected in accordance with the positions the discrete-cosine coefficients take in the discrete-cosine block and may then be output after a plus sign or a minus signal is added to the value selected, and when the discrete-cosine block is subjected to field division and the discrete-cosine coefficients are input in the from of a vertical bit stream, one of the values input from the ten field, compression discrete-cosine transform multipliers to the eight discrete-cosine transform multipliers may be in accordance with the positions the discrete-cosine coefficients take in the discrete-cosine block and may then be output after a plus sign or a minus signal is added to the value selected; and eight adding means associated with the eight selecting means, respectively, each for adding the values output from the associated selecting means. Each of the eight discrete-cosine transform multipliers has, as coefficient, any one of eight inverse discrete-cosine coefficients which are some of the elements of a first matrix applied to perform inverse discrete-cosine transform on the discrete-cosine block and which have absolute values not identical to those of any other elements of the first matrix. Each of the ten field, compression discrete-cosine transform multipliers has, as coefficient, any one of the ten inverse discrete-cosine coefficients which are some of the elements of a second matrix applied to perform field, compression discrete-cosine transform and which have absolute values not identical to those of any other elements of the second matrix.
The inverse discrete-cosine transform apparatus outputs discrete-cosine coefficients multiplied by inverse transform coefficients in the field-mode multiplier, when the input discrete-cosine block is not subjected to field division. The apparatus outputs discrete-cosine coefficients multiplied by inverse transform coefficients in the frame-mode multiplier, when a discrete-cosine block is input in the from of a vertical bit stream and then subjected to field division.
According to the invention, there is provided an inverse discrete-cosine transform apparatus that is designed to perform inverse discrete-cosine transform on a discrete-cosine block that is a matrix composed of at most 8xc3x978 discrete-cosine coefficients. This apparatus comprises: eight multipliers for multiplying the discrete-cosine coefficients input in the form of a bit stream, by coefficients; eight selecting means for receiving the discrete-cosine coefficients multiplied by the coefficients in the eight discrete-cosine transform multipliers; control means for controlling the eight selecting means so that one of the values input from the eight discrete-cosine transform multipliers to the eight selecting means may be selected in accordance with the positions the discrete-cosine coefficients take in the discrete-cosine block and may then be output after a plus sign or a minus signal is added to the value selected; and eight adding means associated with the eight selecting means, respectively, each for adding the values output from the associated selecting means, In the apparatus, each of the eight multipliers has, as coefficient, any one of eight inverse discrete-cosine coefficients which are some of the elements of a matrix applied to perform inverse discrete-cosine transform on the discrete-cosine block and which have absolute values not identical to those of any other elements of the first matrix.
This inverse discrete-cosine transform apparatus effects inverse discrete-cosine transform on a discrete-cosine block that is a matrix composed of at most 8xc3x978 elements.
As can be understood from the foregoing, an inverse discrete-cosine transform apparatus according to the invention has a simple structure. It needs only eight inverse discrete-cosine transform multipliers and only ten field, compression discrete-cosine transform multipliers. This is because standard inverse discrete-cosine transform, compression, inverse discrete-cosine transform, and field, compression, inverse discrete-cosine transform are effected on the inverse transform coefficients of a matrix, thereby extracting the inverse transform coefficients that overlap the others of the matrix in terms of absolute value. The apparatus further comprises eight selecting means, control means and eight adding means. Therefore, it can perform standard inverse discrete-cosine transform, maintaining the resolution of a high- or standard-resolution image subjected to discrete-cosine transform. The apparatus can also effect compression, inverse discrete-cosine transform, converting a high-resolution image subjected to discrete-cosine transform, to a standard-resolution image. Further, the apparatus can execute field, compression, inverse discrete-cosine transform, dividing a discrete-cosine block subjected to field, discrete-cosine transform, into fields, thereby achieving compressed, discrete-cosine transform.
As seen from the foregoing, another inverse discrete-cosine transform apparatus according to the invention has a simple structure. It needs only eight multipliers. This is because standard inverse discrete-cosine transform and compression, inverse discrete-cosine transform are effected on the inverse transform coefficients of a matrix, thereby extracting the inverse transform coefficients that overlap the others of the matrix in terms of absolute value. The apparatus further comprises eight selecting means, control means and eight adding means. Therefore, it can perform standard inverse discrete-cosine transform, maintaining the resolution of a high- or standard-resolution image subjected to discrete-cosine transform. The apparatus can also effect compression, inverse discrete-cosine transform, converting a high-resolution image subjected to discrete-cosine transform, to a standard-resolution image.